fix: update DeepSeek models context window to 128k

[roomote]

This PR attempts to address Issue #7268. Feedback and guidance are welcome.

Summary

Updated DeepSeek models (deepseek-chat and deepseek-reasoner) context window from 64k to 128k tokens to align with the latest DeepSeek API capabilities.

Changes

• Updated context window from 64_000 to 128_000 for both DeepSeek models in packages/types/src/providers/deepseek.ts
• Updated corresponding test expectations in src/api/providers/__tests__/deepseek.spec.ts

Context

According to the DeepSeek API documentation (https://api-docs.deepseek.com/quick_start/pricing/), their models now support 128k context window.

Testing

• ✅ All tests pass
• ✅ Linting checks pass
• ✅ Type checks pass

Fixes #7268

---

Important

Update DeepSeek models' context window to 128k tokens in deepseek.ts and adjust tests in deepseek.spec.ts.

• Behavior:
  • Updated context window from 64_000 to 128_000 tokens for deepseek-chat and deepseek-reasoner in deepseek.ts.
  • Updated test expectations for context window in deepseek.spec.ts.
• Testing:
  • All tests, linting, and type checks pass.

^{This description was created by for a57187b. You can customize this summary. It will automatically update as commits are pushed.}

[roomote]

Reviewing my own code because apparently I trust no one, not even myself.

[roomote]

The context window update to 128k looks correct and aligns with the DeepSeek API documentation. Consider adding a comment referencing the API docs URL (https://api-docs.deepseek.com/quick_start/pricing/) for future maintainers.

Also, this capability update might warrant a changelog entry so users know about the increased context window.

[NaccOll]

@roomote-agent read https://api-docs.deepseek.com/zh-cn/quick_start/pricing and update this pr.

[roomote]

Hi @NaccOll! I see your request to review the DeepSeek pricing documentation and update this PR accordingly. Let me check the documentation and make the necessary updates.

[roomote]

Hi @NaccOll! I have reviewed the DeepSeek pricing documentation at https://api-docs.deepseek.com/zh-cn/quick_start/pricing and updated the PR accordingly.

Changes Made:

1. ✅ Confirmed that both deepseek-chat and deepseek-reasoner models support 128K context window (already correctly updated in the PR)
2. ✅ Updated deepseek-reasoner maxTokens from 8192 to 65536 (64K) to match the documentation which states max output is 64K
3. ✅ Updated the description for deepseek-reasoner to reflect "up to 64K output tokens" instead of "up to 32K tokens"
4. ✅ Added a new test case to verify the correct maxTokens value for deepseek-reasoner

Summary from Documentation:

• deepseek-chat: 128K context, max 8K output ✅
• deepseek-reasoner: 128K context, max 64K output ✅

All tests pass and CI checks are green. The PR now accurately reflects the latest DeepSeek API capabilities.

[daniel-lxs]

LGTM

For reference: https://api-docs.deepseek.com/quick_start/pricing/